[wishingfly]

Both after the first (pre-ampl after RT) and the second (enrichment PCR after tagmentation) PCR you should check your library on the Bioanalyzer. The first time to see you got cDNA and it is worth continuing. The second to see the final library looks fine. Before sequencing I normally check the final library (final pool) on the Bioanalyzer for size and on the Qubit for the concentration (Bioanalyzer is semi-quantitative and not accurate for conc).
In the paper (and during protocol optimisation) I was checking all the libraries on the Bioanalyzer, but just because the numbers were still relatively low. Now that we process hundreds or thousands of cells at the time I check randomly 11 samples for each 384-well plate, get an average conc and use that value to decide how much cDNA I should take for tagmentation. Therefore I used equal volumes for all the samples and not equal amounts. I then do the same for the final library: I run a chip, check the average size and pool the samples using the same volumes for all the samples. It´s not optimal but it is way faster, cheaper and the only feasible way (as far as I know) when you deal with many samples. Drawback: you are pooling also failed/empty wells as well as libraries that were good with others that were very low. Result: you will get huge differences in the number of mapped reads between cells. But, again, I don´t care so much since the cells with few reads can be discarded or the pool resequenced. When working with thousands of cells, even discarding 10% of the cells still give you plenty of information.
Illumina has some normalisation beads which, as far as I can tell from the limited tests I did, work very well. But it is an extra step...and I am trying to get rid of the bead purification steps, not adding new ones!
If your final libraries have a lot of primer dimers that were not removed in the final bead purification, pooling the way I said will generate a lot of reads that have to be discarded, because you will sequence a lot of adaptor dimers (being shorter than your library they will cluster preferentially on the flow cell, of course).

Sounds great, I will preceed with the first round trial next week, following your published protocol and instruction here, thank you so much! Perhaps I might need to bother you for more practical suggestion during the process; I hope I could acknowledge your input if this project could finally get published.

[bagnall.lab]

More Bioanalyzer oddities

Both after the first (pre-ampl after RT) and the second (enrichment PCR after tagmentation) PCR you should check your library on the Bioanalyzer. The first time to see you got cDNA and it is worth continuing.

Hi Simone,
Thank you for all of your valuable insight. It is very helpful. Have you seen anything like the attached bioanalyzer result previously? I thought perhaps contamination but I am surprised it is in all of the samples. Each sample is from 1-3 picked neurons, each from different animals, and some on different days. I ran 22 cycles for the PCR (attempting to compensate for the suboptimal ripping cells from their axons) and thought perhaps it is over-amplification. It also seems that each peak is ~1.8x the size of the previous large peak (295, 527, 938/941, 1728). Any thoughts?
Thank you,
bagnall.lab

[Simone78]

Hi Simone,
Thank you for all of your valuable insight. It is very helpful. Have you seen anything like the attached bioanalyzer result previously? I thought perhaps contamination but I am surprised it is in all of the samples. Each sample is from 1-3 picked neurons, each from different animals, and some on different days. I ran 22 cycles for the PCR (attempting to compensate for the suboptimal ripping cells from their axons) and thought perhaps it is over-amplification. It also seems that each peak is ~1.8x the size of the previous large peak (295, 527, 938/941, 1728). Any thoughts?
Thank you,
bagnall.lab

unfortunately I haven´t seen such a pattern before. We do have sometimes strange peaks but they are mostly only one per sample. In that case it is usually due to a specific transcript that is picked up during the RT and/or PCR (it is an artifact). Have you tried to sequence some of these samples? You should get hundreds of thousands or million reads from the same transcript/region if your case is similar to mine.
Best,
Simone

[wishingfly]

unfortunately I haven´t seen such a pattern before. We do have sometimes strange peaks but they are mostly only one per sample. In that case it is usually due to a specific transcript that is picked up during the RT and/or PCR (it is an artifact). Have you tried to sequence some of these samples? You should get hundreds of thousands or million reads from the same transcript/region if your case is similar to mine.
Best,
Simone

Hi Simone and other colleagues,

I am wondering if you could help me take a quick look at our first trial of cDNA library, with bioanalyzer results attached in the figure. Sample 1-6 are single-cell samples, while sample 7-11 are negative controls (not just water, it is a mimic of the entire cell-pick-up procedure, except picking-up a real cell). I could tell they are not good libraries, and I thought the main problem is RNA degradation during sample harvest; in addition, our facility director suggests the primer-dimers could be another issue. I am wondering what is you opinions on these results.

By the way, here we use 5'-biotin blocked primers, so the TSO concatamers should not be an issue; but do you think we should seek to reduce the primer-dimers? By reducing ISPCR primer concentration? Or do something with beads purification?

Other than trying to better protect our single-cell sample from RNA degradation, do you think increasing preamplification PCR cycles could help? Right now we used 20 cycles; and I think you mentioned earlier you used 22 cycles for T-cells.

Any suggestion is appreciated. Thank you so much!

Best
Alex

[wishingfly]

I have another naive question: for bioanalyzer check, it is really necessary to purify the DNA after PCR preamplification? I got the answer "yes" from both Aligent tech support and our facility director; however, it seems someone in this threads series did mention they did not purify the PCR product before bioanalyzer. It is not a big issue, but I just want to save some time (and beads) while optimizing the protocol. Thanks a lot!

[nucacidhunter]

I have another naive question: for bioanalyzer check, it is really necessary to purify the DNA after PCR preamplification? I got the answer "yes" from both Aligent tech support and our facility director; however, it seems someone in this threads series did mention they did not purify the PCR product before bioanalyzer. It is not a big issue, but I just want to save some time (and beads) while optimizing the protocol. Thanks a lot!

Without clean up step the fragments will not run in correct size and batch to batch comparison would be affected. Bioanalyzer assays designed to run with samples resuspended in TE buffer but any other resuspension buffer should be fine as long as one takes that into account when interpreting results. Reaction samples would have salts and proteins and they will interfere or interact with DNA fragments during run. If you are going to analyse lots of samples, you can run few samples before and after clean up to get an idea of reaction buffers and enzymes effect. Cleaned samples should be eluted or resuspended in the same volume of reaction minus 10-15% to maintain DNA concentration in both before and after clean up samples.

[Simone78]

I have another naive question: for bioanalyzer check, it is really necessary to purify the DNA after PCR preamplification? I got the answer "yes" from both Aligent tech support and our facility director; however, it seems someone in this threads series did mention they did not purify the PCR product before bioanalyzer. It is not a big issue, but I just want to save some time (and beads) while optimizing the protocol. Thanks a lot!

for a qualitative check it is perfectly fine to run a chip straight after pre-amplification (dilution with water is not necessary). I do it all the time. I don´t have an example file with me but I can send it to you on Monday. The only problem with this approach is that sometimes you see quite some primer dimers that, if you would then proceed to the tagmentation step without purification, might lower the quality of your data (they also get tagmented).

[Simone78]

Hi Simone and other colleagues,

I am wondering if you could help me take a quick look at our first trial of cDNA library, with bioanalyzer results attached in the figure. Sample 1-6 are single-cell samples, while sample 7-11 are negative controls (not just water, it is a mimic of the entire cell-pick-up procedure, except picking-up a real cell). I could tell they are not good libraries, and I thought the main problem is RNA degradation during sample harvest; in addition, our facility director suggests the primer-dimers could be another issue. I am wondering what is you opinions on these results.

By the way, here we use 5'-biotin blocked primers, so the TSO concatamers should not be an issue; but do you think we should seek to reduce the primer-dimers? By reducing ISPCR primer concentration? Or do something with beads purification?

Other than trying to better protect our single-cell sample from RNA degradation, do you think increasing preamplification PCR cycles could help? Right now we used 20 cycles; and I think you mentioned earlier you used 22 cycles for T-cells.

Any suggestion is appreciated. Thank you so much!

Best
Alex

Indeed it looks degraded. Increasing the number of cycles won´t help. there must be something else during the sample preparation and before the sorting/picking that degrades the cell. It might also be that you have some kind of contamination going on. My negative controls (just water instead of the cell) look flat, except from some (unused) primers that are not removed with the bead purification.
How long does it take to get the cells ready for picking? Do you freeze the cells immediately after picking? Do you work in a clean environment (hood/clean room)? Have you tried to run a positive control (10-20 pg of good-quality tot RNA)? If also the RNA looks like that, then the problem is somehow in how you handle the samples (or in some reagents). I hope you can find the source of the problem!

[wishingfly]

Indeed it looks degraded. Increasing the number of cycles won´t help. there must be something else during the sample preparation and before the sorting/picking that degrades the cell. It might also be that you have some kind of contamination going on. My negative controls (just water instead of the cell) look flat, except from some (unused) primers that are not removed with the bead purification.
How long does it take to get the cells ready for picking? Do you freeze the cells immediately after picking? Do you work in a clean environment (hood/clean room)? Have you tried to run a positive control (10-20 pg of good-quality tot RNA)? If also the RNA looks like that, then the problem is somehow in how you handle the samples (or in some reagents). I hope you can find the source of the problem!

Thank you for your suggestion. I agree with you that the RNA degradation is the major issue we should overcome in the next step; we did have positive controls, but the samples will be processed and subject to bioanalyzer today, on Friday the equipment was so busy that we only have time for 1 chip.

Based on the current result, do you think the primer dimers is an issue as well? Would you mind posting your bioanalyzer traces before and after purification, so that we could get an idea of what does it looks like when primer dimers dominate? Thanks a lot!

[bplevi]

Hi Simone and other colleagues,

I am wondering if you could help me take a quick look at our first trial of cDNA library, with bioanalyzer results attached in the figure. Sample 1-6 are single-cell samples, while sample 7-11 are negative controls (not just water, it is a mimic of the entire cell-pick-up procedure, except picking-up a real cell). I could tell they are not good libraries, and I thought the main problem is RNA degradation during sample harvest; in addition, our facility director suggests the primer-dimers could be another issue. I am wondering what is you opinions on these results.

By the way, here we use 5'-biotin blocked primers, so the TSO concatamers should not be an issue; but do you think we should seek to reduce the primer-dimers? By reducing ISPCR primer concentration? Or do something with beads purification?

Other than trying to better protect our single-cell sample from RNA degradation, do you think increasing preamplification PCR cycles could help? Right now we used 20 cycles; and I think you mentioned earlier you used 22 cycles for T-cells.

Any suggestion is appreciated. Thank you so much!

Best
Alex

Hi Alex and Bagnall.lab,

We have been doing SmartSeq2 with good results for about a year. Then about two months ago our amplified cDNAs started looking terrible. Our background looks exact same as yours (see attachment). We’ve spent the last month or so trying to sort out the problem, and in the process have replaced every single reagent about 2 times now. We believe the contamination is bacterial RNA in SuperScriptII. The reason we believe this is: 1) those peaks are reproducible and show up in water only samples (quite strongly), 2) we do not see those peaks if we omit the SuperScript2 enzyme (so they aren’t an amplifiable DNA contaminant), 3) we do not see any background peaks if we use an old aliquot of SuperScript2 (see attachment) and 4) we don’t see those peaks if we use SuperScript3 or ProtoScriptII. We have also sequenced a water control on the MiSeq and the top reads map to bacteria.

We have spent a fair amount of time talking to Life Tech about this problem, have tried several new batches of enzymes they sent to us, but all had this background contamination. So our solution has been to go with ProtoScriptII from NEB. I don’t like switching enzymes but this was the only way we were able to get rid of the contamination and maintain consistent-looking amplified cDNA. We looked at SuperScript3, but its amplified cDNA looks quite a bit different from SuperScript2 cDNA with one very prominent peak. ProtoScriptII amplified cDNA looks very much like SuperScript2 libraries, and we will compare the two enzymes using standard control RNA (don’t have this data yet).

This problem was a serious setback for us and we lost precious samples and time getting through it. I’d be interested in knowing if others have experienced this problem and what their work around was.

Good luck.

[wishingfly]

Hi Alex and Bagnall.lab,

We have been doing SmartSeq2 with good results for about a year. Then about two months ago our amplified cDNAs started looking terrible. Our background looks exact same as yours (see attachment). We’ve spent the last month or so trying to sort out the problem, and in the process have replaced every single reagent about 2 times now. We believe the contamination is bacterial RNA in SuperScriptII. The reason we believe this is: 1) those peaks are reproducible and show up in water only samples (quite strongly), 2) we do not see those peaks if we omit the SuperScript2 enzyme (so they aren’t an amplifiable DNA contaminant), 3) we do not see any background peaks if we use an old aliquot of SuperScript2 (see attachment) and 4) we don’t see those peaks if we use SuperScript3 or ProtoScriptII. We have also sequenced a water control on the MiSeq and the top reads map to bacteria.

We have spent a fair amount of time talking to Life Tech about this problem, have tried several new batches of enzymes they sent to us, but all had this background contamination. So our solution has been to go with ProtoScriptII from NEB. I don’t like switching enzymes but this was the only way we were able to get rid of the contamination and maintain consistent-looking amplified cDNA. We looked at SuperScript3, but its amplified cDNA looks quite a bit different from SuperScript2 cDNA with one very prominent peak. ProtoScriptII amplified cDNA looks very much like SuperScript2 libraries, and we will compare the two enzymes using standard control RNA (don’t have this data yet).

This problem was a serious setback for us and we lost precious samples and time getting through it. I’d be interested in knowing if others have experienced this problem and what their work around was.

Good luck.

Hi bplevi,

Thank you for your timely input! I was astonished that I did not expect the problem could be the contamination of RTase. In my first trial, the negative controls are not pure water, rather they are the pipet solution that mimic exactly the same procedure of cell harvest, so originally I thought maybe I picked up some free RNA in the tissue culture media. Now it seems the contamination could come from the enzyme. Today I did another batch of negative controls with pure water, and fortunately we also have some old SuperScript II, so I did a parallel compare between old vs. new enzyme; I will keep you updated when the bioanalyzer results are ready.

As to the alternative of SuperScript II, if we don't trust it any longer, do you suggest using ProtoScript II from NEB? It sounds like you are going to try, but have not done yet? Have you considered switching to SuperScript IV? Simone mentioned in previous threads it works fine, so I am thinking if SuperScript II is no longer reliable, whether we should switch to other options right now.

[bagnall.lab]

Wow bplevi, thanks for the tip. That's pretty disturbing. I'm glad you followed it out as far as the sequencing, that really nails it. We've been using SS II and SS IV, and are getting possibly similar peaks with both, so we may want to try the NEB version instead...

[Simone78]

Thank you for your suggestion. I agree with you that the RNA degradation is the major issue we should overcome in the next step; we did have positive controls, but the samples will be processed and subject to bioanalyzer today, on Friday the equipment was so busy that we only have time for 1 chip.

Based on the current result, do you think the primer dimers is an issue as well? Would you mind posting your bioanalyzer traces before and after purification, so that we could get an idea of what does it looks like when primer dimers dominate? Thanks a lot!

Please find attached a test (using tot RNA, not cells) of the difference between pre-amplified cDNA before and after bead purification. Samples were either purified manually or with a robot (even if it didn´t go very well in this case, for some reasons).
Now the dilemma is the following: skipping bead purification doesn´t remove the leftover adaptors, that can thus be tagmented (they will) and end up in your final library and then in your final data.
On the other hand, doing bead purification gives a nice and clean prep but you will go crazy with the purifications, especially if you have thousands of samples to process...and even if you use a liquid handling robot (I do, right now).

In conclusion, I tend to prefer the first solution (no purification).
If you then want more data from each cell just pool less or resequence the pool. But that´s just an idea, I haven´t really done it yet.

[wishingfly]

Hi everyone,

I would like to update my evidence of possible contamination of SuperScript II from Invitrogen, as you can see from the attached figure, we compared the new order (placed in May 2015) vs. an old stock (ordered years ago); the only difference is the enzyme by itself, we use the same (new) first strand buffer, DTT and RNaseOUT. We have 3 replicates in each condition; negative controls are RNA free water, and positive controls are 10 pg standard RNA. Considering other users, such as "bplevi", also report exactly the same problem, we have reason to believe the pipeline of this product from Invitrogen is contaminated, and it is unlikely to be resolved by simply order a new pack.

So I think we should seek for alternatives. "bplevi" suggested the Protoscript II from NEB; if any of you tried it out, could you give us some feedback on how it works? How about other options, such as SuperScript IV?

Thanks a lot!
Alex

[Simone78]

Which is the lot number of the "bad" SSRT II? We are running out of the old one and now I´m a bit concerned...

btw, Protoscript II worked ok in my hands (not impressed, though), but Superscript II was clearly better. As alternative I would suggest to use PrimeScript from Clontech, also a MMLV-based RT.